Re: Insertions

> If your output is serialised text, this sidesteps some of the namespace issue.
> 
> But if your output is XDM nodes, 


I think we might need to talk about what the output of a conforming iXML processor is. 

From the spec:

> "An input is parsed using this grammar, and the resulting parse tree is serialised as XML.”
> 
> "If the parse succeeds, the resulting parse-tree is serialised as XML by serialising the root node.”
> 
> "If more than one parse tree describes the input, the processor must serialize one of them."

Throughout the spec, we talk about serialized XML as the output, but then in the section “Conformance of processors”, we say this:

> "The form in which XML documents are produced is not constrained by this specification; processors should be capable of producing serialized XML as a character stream, but other forms (e.g. DOM instances or XDM instances) may also be used.”


I think we need to be clearer about what, exactly, the ixml processor will produce. Because currently, the main wording of the spec does appear to constrain the form of the vXML output by stating that serialized XML will/must be produced.

We require an output document in the case of failure, so perhaps it makes sense to require one in the case of success?

Very best,

Bethan

___________________________________________________ 
Dr. Bethan Tovey-Walsh 
Myfyrwraig PhD | PhD Student CorCenCC 
Prifysgol Abertawe | Swansea University 
Croeso i chi ysgrifennu ataf yn y Gymraeg.

> On 4 Mar 2022, at 10:49, Tom Hillman <tom@expertml.com> wrote:
> 
> data: @xmlns, value+",".
> xmlns: +"http://example.com/data".
> If your output is serialised text, this sidesteps some of the namespace issue.
> 
> But if your output is XDM nodes, this should throw well-formed-ness errors as namespace declarations are not attributes, and attribute names starting with [xX][mM][lL] are forbidden (as I understand it).
> 
> Perhaps a mark other than '@' for namespace declarations would be an acceptable compromise here?
> But a mechanism that allows for a default namespace binding and no other seems
> pretty limiting and arbitrary to me.
> 
> It's only limited to the default namespace binding if colons are disallowed in rule names, right?
> 
> _________________
> Tomos Hillman
> eXpertML Ltd
> +44 7793 242058
> On 3 Mar 2022, 17:51 +0000, Norm Tovey-Walsh <norm@saxonica.com>, wrote:
>> Steven Pemberton <steven.pemberton@cwi.nl> writes:
>>> Some time ago Tom suggested using +"abc" to signal a textual insertion
>>> into the serialisation, which we put on a backburner. But I ended up
>>> thinking about it today anyway.
>> 
>> It’s nice. On the whole, I think I prefer “^” to “+”, but the bigger
>> question, I think, is whether not we should be taking this up now.
>> 
>> I’d much rather get pragmas sorted out and bedded in before we take on
>> more new work. I also think this example:
>> 
>>> data: @xmlns, value+",".
>>> xmlns: +"http://example.com/data".
>> 
>> will only fuel more debate about qualified names in V1. “There are no
>> qualified names in vxml” is a reasonable position, I think. But a
>> mechanism that allows for a default namespace binding and no other seems
>> pretty limiting and arbitrary to me.
>> 
>> Be seeing you,
>> norm
>> 
>> --
>> Norm Tovey-Walsh
>> Saxonica